SVM Based Neuro Fuzzy Model for Short Term Wind Power Forecasting

Wind power forecasting is the major area of concern in wind power generation due to the unpredictability of wind speed. Existing soft computing and statistical methods focus towards deterministic wind forecasting and neglects the uncertainties associated with wind flow. In this paper, the wavelet decomposition combined with adaptive neuro fuzzy inference system is used for short term wind power forecasting. Wavelet decomposition handles the wind power data series fluctuations and the neuro fuzzy system ensures fast learning in a real time scenario. Support vector machine (SVM) is utilized to reduce the prediction errors. A quantile regression has been applied on the SVM output to improve accuracy. The proposed forecasting method is assessed using the real time data collected from the Coimbatore wind farm, India. The assessment is conducted in different seasons to reduce the errors and improve the prediction accuracy.